Longreads + Open Thread

Longreads

• Derek Thompson and Ryan Burge on the Substackification of American religion. This is a fascinating piece: there are periodic trend stories about the revival of Christianity among young people, and it's easy to find anecdotal evidence. But it's very hard to find statistical evidence of this (partly because young people are hard to survey in general). This piece has some fun narrative violations—the growing part of religious adherence is non-denominational Christianity, not more traditional churches, for example. And where it gets really interesting is in exploring how that phenomenon plays out: when a church isn't part of an organized hierarchy, it's going to be organized around one charismatic leader, and when that leader retires or goes through some kind of scandal, there isn't an institution to smooth things over. But you can rewind that and say that every organized religion handled the transition from a prophet or messiah figure to a bureaucracy.
• The other day, I stumbled on a fun question on Twitter: how did college students "do the reading" when books were much more expensive? Some clicking around revealed this fascinating paper on that very topic. As it turns out, the concept of a textbook is descended from class notes, pirated sermons, and other informal spoken-to-written transitions. So, in a way, the people who's browser autocomplete doesn't fill in "linkedin.com" when they type "li" into the address bar are part of an ancient and honorable tradition, one that includes many of the people responsible for the primary sources they read.
• There is a bug in the Apollo Guidance Computer, which could have, hypothetically, have forced Apollo astronauts to reset their guidance computer. (Edit: as it turns out, this bug was both less catastrophic than it looked and eventually identified by NASA.) Which is not the end of the world. Mission Control would presumably get to "can you turn it off, then turn it on again?" before the crew ran out of oxygen. Still, it's likely that an issue like this could have set the development of computers back by years, or more, and it was a reasonable outcome that such a bug could happen. This piece is also a fun example of the history parallax effect of AI: everything that can be figured out based on disparate clues scattered across different documents will be figured out, in the next few years.
• After their defeat in the Second World War, Japan was able to rebuild their economy, and in many areas surpass the US; for a while in the 1980s, the popular media narrative was that eventually, Japan would own everything. This piece was written within a year of the Nikkei's peak and thus the beginning of Japan's long deflationary period. It's all about how people in Japan are willing to tolerate an expensive, inefficient domestic economy in exchange for a more competitive export-oriented one. But in retrospect, it was written at the peak of the average Japanese person's willingness to tolerate a system where Japan had a high GP per capita and they had a lower-than-average quality of life. 
• Dwarkesh interviews Michael Nielsen, mostly on the history of and nature of science. Many great points from both of them. One fun thread is that when we talk about falsifying theories, what we generally mean is that in a family of explanations, we can run experiments that disprove one or more of them, but that only leads to scientific progress when we disprove all but one, and only until the next round of more comprehensive models: some of the experimental data that led to special relativity came from experiments to determine the nature of the ether. There's also some great speculation about how differently technology could have developed. People sometimes borrow the term "tech tree" from video games: in games, you'll have a list of technologies that are dependent on prior technologies (so researching ships that can sail across the ocean might depend on researching sails and the compass, for example). That metaphor tends to imply that there's one deterministic way to get where we are. But the other way to think about it is that there are many possible tech trees, and that when we choose one, we're missing out on developments further out in other tech trees: maybe there's some amazing invention downstream from DDT or thalidomide that a different version of humanity, or an alien civilization, might discover. And, were we to encounter such a civilization, they might be blown away by some invention we made that was downstream from legalized sports betting or gas-powered leaf blowers or some other tool that might hypothetically have been banned as a nuisance.
• This week's Capital Gains is another financial history piece, this time on what we can learn from the Ford IPO, which was at the time the largest offering in history. It's illuminating to look at all the ways that IPO differs from upcoming ones like SpaceX, Anthropic, and OpenAI. But some things never change.
• A great Read.Haus question: why do people do barbell portfolios instead of trading the entire curve? In asset allocation, what you'll often find is that if you follow modern portfolio theory and try to backtest your way into a good portfolio, what you end up with is a portfolio that's a mix of one mostly safe asset class and one risky asset class, in whatever ratio gets you to your target volatility. The identity of these assets depends on the window of your backtest: if you do it now, your risk asset is probably US growth equities; at various points in the past it might have been emerging market equities, PE, VC, or something else. If different asset classes have basically reasonable risk/reward tradeoffs, then your historical analysis will end up showing some of them as slightly better and others slightly worse than their long-term returns simply because there's some variance in performance over time and it's hard to set the right timeframe for backtests. (Does your backtest for venture investing include equity in the various East India companies? Does it start when venture got formalized as an asset class, or does it start a little earlier when nobody had defined venture capital well and young, growing companies were absolutely starved for capital? Does your PE backtest include the time the Du Pont family did an LBO on their own business, or does it start with KKR—and if it starts with them, what's the justification for omitting the earlier deals that KKR's principals did while they were at Bear Stearns?) So the reason people don't trade the whole curve is that even in finance, where you can talk coherently about the historical returns from different allocation strategies, we have to fudge things a bit to get a sensible mix. There's no way to backtest different splits of time between working at your day job, raising your kids, and working on your novel. And the time you spend musing on this probably comes out of the free time that you could otherwise spend on the most meaningful or highest-variance use of time. So in this case, the way to apply financial reasoning to life is to say that the assumptions behind that financial reasoning don't apply, and you have to take the occasional outright gamble.

Open Thread

• Drop in any links or comments of interest to Diff readers.
• On the topic of the first link, are there any other interesting, counterintuitive details on how teaching has evolved over time in response to changes in information technology?

Diff Jobs

Companies in the Diff network are actively looking for talent. See a sampling of current open roles below:

• Well-funded, frontier AI neolab working on video pretraining and computer action models as the path to general intelligence is looking for researchers who are excited about creating machines that learn from experience, not text. Ideally you have zero-to-one pre-training experience and/or are a high-slope generalist who’s frustrated that the big labs aren't doing this. (SF)
• Series A startup building multi-agent simulations to predict the behavior of hard to sample human populations is looking for researchers and engineers (ML, platform, infrastructure, etc.) to improve simulation fidelity and scale the platform to hundreds of millions of simulation requests. Problem-solving and genuine interest in simulation matter more than pedigree. Experience working with languages with an algebraic type system is a plus. (NYC)
• A Fortune 500 cybersecurity company with decades of proprietary security data is running an internal incubation with a pre-seed startup mentality and a mandate to build something new in AI. They are looking for a founding engineer who can ship fast, an engineer with a security background who’d be excited to contribute to OpenClaw’s security efforts, an AI researcher, and a generalist (ex-banking/consulting/PE background preferred) who wants to wear a bunch of different hats. Comp is FAANG+ and cash heavy. If you want to build something new in AI, but also need runway, this is for you. (SF/Peninsula) 
• High-growth startup building dev tools that help highly technical organizations autonomously test and debug complex codebases is looking for senior product managers who enjoy defining developer-facing APIs and abstractions. Experience with fuzzing or property-based testing a plus! (London, D.C.)
• A leading AI transformation & PE investment firm (think private equity meets Palantir) that’s been focused on investing in and transforming businesses with AI long before ChatGPT (100+ successful portfolio company AI transformations since 2019) is hiring experienced forward deployed AI engineers to design, implement, test, and maintain cutting edge AI products that solve complex problems in a variety of sector areas. If you have 3+ years of experience across the development lifecycle and enjoy working with clients to solve concrete problems please reach out. Experience managing engineering teams is a plus. 

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.